基于PCNN和DSP的感興趣區域提取系統研究與實現

摘 要:由于普通PC實現的感興趣區域提取系統普遍存在耗時大、實時性差且處理精度不高等缺陷，在此利用瑞泰公司ICETEK-DM642-PCI評估板提供的TI公司的TMS320DM642芯片，DSP/BIOS的實時操作系統，CCS集成開發環境來構建應用程序。應用脈沖耦合神經網絡對圖像進行噪聲抑制，然后進行二值分割。再利用PCNN正向自動波去除一些很小的干擾，反向自動波恢復感興趣的區域并提取出。實驗結果表明，采用的算法能夠精確地提取出圖像中感興趣的區域。

關鍵詞: DM642; 脈沖耦合神經網絡; 感興趣區域提取; 二值分割

中圖分類號:TN911.7; DM642 文獻標識碼:A

文章編號:1004-373X(2010)14-0144-04

Research and Implementation of Interested Region Extraction System Based on PCNN and DSP

ZHAO Guang-peng， DUAN Zhong-xing， DING You-jun

(School of Information and Control Engineering， Xi’an University of Architecture and Technology， Xi’an 710055， China)

Abstract: Because the interested region extraction algorithm based on common PC has the defects such as time-comsumption， bad real-time and poor processing accuracy， the application program is built with TI's TMS320DM642 chip provided by ICETEK-DM642-PCI evaluation board of Rite-Hite Corporation and DSP/BIOS real-time operating system under the integrated development environment of CCS， the noise suppression of image is performed by the pulse-coupled neural network， two-value segmentation is carried out， some small interference is removed by the foreward automatic wave of PCNN， the interested region is recovered by the backward automatic wave of PCNN， and then the interested region is extracted. Experimental results show that the algorithm can be used to extract the interested region of images accurately.

Keywords: DM642; pulse-coupled neural network; extraction of interestrd region; two-value segmentation

0 引 言

在圖像數據中包含著大量而繁復的信息，然而在實際應用過程中往往需要提取出感興趣的區域。例如在醫學領域，對于一些醫用藥品的檢測和細胞的識別就需要能夠快速而準確地提取出感興趣的區域。普通PC機實現的感興趣區域提取系統普遍存在耗時大、實時性差且處理精度不高等缺陷，是因為計算機的硬件架構和外圍設備存在著瓶頸，無法提升。而專門為圖像處理設計的高速數字信號處理芯片 DM642可以在一個時鐘周期內完成一次乘操作，其處理速度要遠遠快于普通的PC計算機，顯然更適合用于需要做大量重復乘操作的圖像處理。以往的感興趣區域提取算法都是只對于圖像進行識別，經常會出現對區域提取信息的丟失，或者精度不高。而本文給出的PCNN具有正向消除干擾和反向修復圖像的能力，在精確性方面要優于其他算法。……